FORUM

On the First Law of Geography: A Reply

D
aniel Sui has sent me a written version of Miller calls my simulation effort ‘‘crude,’’ whereas Barnes
comments presented by five geographers at a considers that it invokes complex ideas. Miller would
panel on the first law of geography organized undoubtedly consider economics, transportation, geol-
by him at the 2003 AAG meeting in New Orleans. The ogy, and other factors to produce a much richer, but also
comments seem to fall into two camps: some reject more involved, model. Curiously there is no discussion of
the idea of ‘‘laws’’ in geography, and others feel that my other concepts from the 1970 paper.
notion has been of some merit. Interestingly, several
other laws are cited in the comments; two by Isaac On Laws
Newton (gravity, motion), two additional ‘‘first’’ laws
(ecology, social science), four additional laws applying Is the first law of geography simply an observation
to people (utility maximization, primate city, human better referred to as an observed regularity or a principle
behavior, and Gresham’s), and three ‘‘second’’ laws and best not called a law? Semantics aside, how can one
(thermodynamics, spatial heterogeneity, things are never tell when something qualifies as a law? So what makes an
equal—the last two being only suggestions). Curiously, observation a law? Two of the discussants on the panel
they do not mention other well-known laws, such as placed great emphasis on this question. One, like many
those of Zipf (1949), or Ravenstein’s (1885) 10 laws, or a social scientists, believes that there are not, and cannot be,
second law of geography by proposed by Arbia, Benedetti, any such thing as laws where human behavior is involved.
and Espa (1996) and one by myself (Tobler 1999, 87).1 One argument is that once an empirical regularity—per-
The comments then focus on three topics. One is a haps qualifying for the status of a law—is discovered, it will
discussion of what constitutes a law, and whether ‘‘the be modified by the people involved and thus rendered
first law of geography’’ fits into the appropriate defini- inoperable. Another strongly held view is that looking for
tion. Much of the remaining material examines the laws of human behavior is a misguided effort involving
concepts of ‘‘related’’ and ‘‘near.’’ Smith is quite correct ‘‘physics envy.’’ This approach argues that social science
in pointing out that this discussion would never have should be action oriented, and that the purpose is to
taken place if the specific word law had not been used. change the world, or at least society. Many contemporary
And Barnes is also correct when he puts it into context writers on human behavior represent this attitude, in-
and observes that, by restricting myself to local effects, cluding Flyvbjerg and Giddens to whom I return later.
I used the notion to parse the possible complexities of Smith points out that his stance is one of faith, not
simulating urban growth. refutable by any evidence (‘‘beyond validation or falsi-
I am a great believer in simplicity, when this is pos- fication’’), and he uses philosophical citations to argue
sible. For example, the point in science is to achieve as that ‘‘the first law’’ is not a law at all. I would counter
many results as possible with the fewest hypotheses. So, this by observing that scientists have demonstrated that
in order to simplify the problem of depicting the growth the pronouncements of philosophers are often suspect
of population in the Detroit region, I tried to eliminate and incorrect (Reichenbach 1952; Lakatos and Feyer-
complicating factors. This is when I invoked ‘‘the first abend 1999, 19–112). A mathematician (Simmons 1992,
law of geography: everything is related to everything else 150) is even more severe, writing:
but near things are more related than distant things.’’
Doing this allowed me to concentrate on local effects— The growth of empiricism and the rise of science over the
using the idea of a change in the ‘‘unit inhabitant,’’ and past three centuries have made it almost impossible to
ignoring many other possible influences. As a result, take seriously the extravagant pretensions of the a priori

Annals of the Association of American Geographers, 94(2), 2004, pp. 304–310 r 2004 by Association of American Geographers
Initial submission, July 2003; final acceptance, November 2003
Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, and 9600 Garsington Road, Oxford OX4 2DQ, U.K.
 Tobler: On the First Law of Geography: A Reply 305

philosopher, who sits in his study and spins a web of words, social science and cites Giddens (1982) who later writes
fanciful imaginings, and empty speculations out of the that it ‘‘is preferable not to use the term [law] in social
material of his own consciousness. Faith in reason alone is science’’ (Giddens 1984, 347).2 In some ways Giddens’
alien to us, and we believe that only careful observation entire oeuvre emphases this theme. I do not find these
and experiment can reveal anything of substance about the pronouncements very creditable. Just as in previous
actual universe. centuries it was widely believed that humans could never
Kant is the classic example. His ideas concerning travel to the moon, such assertions are not proofs.
geometry were completely wrong as demonstrated by Miller makes the further point that Newton’s law of
the invention of non-Euclidean geometries. Therefore gravity does not explain the phenomena, it just describes
a preferred—by me—definition of a law comes from a the effect. His laws of motion do the same. Thus, re-
scientist, namely Richard Feynman (1967). In his book ference to the need for an explanation, or a ‘‘cause,’’ is
on The Character of Physical Law, he describes how to not a useful argument, but more of a human psycho-
invent a new law. He points out that the first, and most logical requirement. People seem to have a need to find
difficult, step is to guess (Feynman 1967, 156). Then the ‘‘causes,’’ and, therefore, laws about human behavior
criterion becomes to ‘‘compute the consequences of that only describe, but do not explain, are regarded as
the guess to see what would be implied . . . . Then we being inadequate.
compare the result . . . with experiment or experience In my reply to the discussants, I seem to have been
. . . [or] observation to see if it works. If it disagrees with forced to defend the first law of geography. Consequently,
experiment it is wrong.’’ So guess again. He also points I looked around for several additional laws pertaining to
out that laws can only be discovered by doing something human behavior. These include several other references
radically different. The procedure used by previous dis- to ‘‘laws’’ in the social sciences. For example, Zipf’s several
coverers of laws, therefore, cannot succeed. Feynman laws based on his principle of least effort—one on spatial
suggests that there is no fixed method that can lead to interaction and another on word frequencies—are often
the discovery of laws. cited, as is Pareto’s law in economics relating to the dis-
Ravenstein’s (1885) ‘‘Laws of Migration,’’ although tribution of incomes. The Auerbach-Pareto-Zipf law of
not given in a formal form, do allow prediction as city sizes is analyzed by Mandelbrot (1965). He also
pointed out by Dorigo and Tobler (1983)—in another mentions (1965, 322) an Estoup-Zipf law of word fre-
laws’ paper. Ravenstein, referring to migration between quencies. Lotka (1929) proposed a law concerning the
counties in Great Britain, wrote, frequency distribution of scientific productivity; Brad-
ford’s law (Garfield 1980) is similar. Also in economics
even in the case of ‘‘counties of dispersion’’ which have a
there is Say’s (1803) ‘‘law of markets: supply creates its
population to spare for other counties, there takes place an
own demand’’ (Patinkin 1948, cited in Weintraub 2002,
inflow of migrants across that border which lies furthest
away from the great centres of absorption.
159). In sociology Merton (1973, 16) mentions Scheler’s
—(1885, 191) ‘‘law of three phases,’’ and Thorndike’s ‘‘law of effect’’
comes from psychology (cited in Lewin 1951, 27, 66). The
Consider the case of Mexico, for which the center of Stanford nonmetric scaling pioneer psychologist Roger
absorption is the United States. Mexico, as predicted Shephard has a law attributed to him in relation to
by Ravenstein, has an (illegal) immigrant problem, with multidimensional scaling: ‘‘if a solution exists, probably it
people coming in from its neighbor to the south. This exists in two dimensions’’ (Coxon 1982, 87). The French
is clearly a correct prediction from the ‘‘migration law.’’ engineer Lalanne in 1875 proposed ‘‘la loi de l’équilatéralie’’
This makes the case for the justification of the word law and ‘‘la loi des distances multiples,’’ having to do with the
by Ravenstein, using the Feynman criterion. distribution of towns and route connections, anticipating
As another example, in Tobler and Weinberg (1971), Christaller by half a century (Palsky 1996, 103). There is
the spatial gravity model is inverted to make a prediction also a ‘‘first law of cognitive geography’’ (Fabrikant et al.
of the location of ancient Hittite villages in central 2002). But my favorite discovery by far is David Lodge’s
Turkey. Is not the empirical efficacy of the widely used ‘‘law of academic life: it is impossible to be excessive in
gravity model of human interaction, whether in its flattery of one’s own peers’’ (Lodge 1984, 152; emphasis as
simplest form or as an entropy model, a sufficient dem- in the original—cited in Sokal and Bricmont 1998, 259).
onstration of the validity of the first law of geography? I have pointed out that in social science some laws do
In Flyvbjerg (2001, 44–45) we find an assertion that exist, or at least that some regularities have been called
the social sciences are incapable of producing laws. He laws. A longer discussion about laws in geography can be
advocates an action-oriented, invariably interventionist, found in Golledge and Amadeo (1968). Also see Hempel
306 Forum: On Tobler’s First Law of Geography

(1996) who provides pro and con arguments for the field ellipsoidal) geodesic (as the crow flies) distance there are
of history. History is often cited as being similar to ge- distances in units of time, or travel cost, or intervening
ography in its formulation as an academic endeavor. opportunities, city block distance (the so-called Man-
hattan metric), and various Riemannian or Finsler dis-
Everything Is Related tances. Some speak of social distances, time-varying
network distances, topological distances, genetic dis-
The phrase ‘‘everything is related to everything else’’ tances, ordinal distances (far, farther, farthest), and so
in my paper was sufficiently vague to have resulted in on. Often these distances are not symmetric. We refer
much discussion. I don’t think that it is literally true, but to the friction of distance. Thus, proximity and near can
it is frequently asserted, as evidenced by the example of take on many meanings in different situations.
Lorenz’s Brazilian butterfly, cited in Phillip’s commen- One interpretation, incorporating both relations and
tary. In a contemporary advertising flyer, The Economist, nearness, is the auto-covariance (or varigram) function,
a British periodical, uses the butterfly theory to en- known for at least 40 years (Gandin 1963). This, along
courage you to subscribe to their publication in order to with Kriging and interpolation, seem to be the context in
stay on top of such ‘‘seemingly insignificant or remote which the first law is most often cited, as evidenced by
events.’’ It again occurs in the ‘‘first law of ecology,’’ also an examination of the phrase as referenced by Google on
mentioned by Phillips. We do know that an event in the Internet. In migration studies I have found strong
Sarajevo had a profound influence throughout the world. spatial adjacency correlation of effects in movement
Cannot this be used to justify the study of geography for patterns (Tobler 1995). The angle of repose in dry sand
students and others? results in a phenomenon that is quite similar to spatial
My former colleague Jack Estes, in speech a few years decay in social interaction as observed by Hägerstrand
ago (Estes 1996) referred to the conservationist John (1957), although the reason for the two is very different.
Muir and quoted him as saying, ‘‘When we pick out Multidimensional scaling, a frequently used technique
anything by itself we find it hitched to everything else in in the social sciences, explicitly makes the assumption
the universe.’’ Estes then quotes from Francis Thomp- that similarity implies nearness (Goodchild and Janelle
son’s poem, ‘‘Mistress of Passion,’’ as follows: 1988). Seriation methods used in archaeology, with re-
spect to closeness in time, and ordinations in political
All things by a mortal power near or far,
science and psychology invoke related ideas (Hubert
hiddenly to each other linked are,
that thou canst not stir a flower without the troubling of 1974). Both the archaeologist and the historian have as
a star. one of their central dogmas the notion that the present
is influenced by the past. There is a similar relation be-
And ‘‘This we know: All things are connected’’ is tween events in space. Geographers are dogmatic that
a statement about life in a poem attributed to Chief what happens at one location is influenced by events at
Seattle by J. Cameron (1997, 38). A similar quotation other places. This relation between events and places
from Chief Seattle’s 1854 letter to the U.S. government allows one to make spatial predictions. For clarification
is found in the works of Joseph Campbell (1988, Program this may be compared with temporal prediction.
4, Introduction).3 The ‘‘law of unexpected consequences’’ The method of radiocarbon dating assumes that the
has a similar theme.4 degree of similarity of samples can be translated into a
In light of these quotes, my statement does not appear time difference to predict a date. Glottochronology—the
to be very original; perhaps it is best to consider it poetic study of language change over time—has a similar ob-
license. But it reinforces the first law of geography by jective and uses a comparable strategy. It is occasionally
putting it into a wider context. All I did was to add the used to estimate how long ago two culture groups sepa-
geographic insight. An analogy can be made to con- rated. In these studies we can recognize the assumption
vergent infinite series where subsequent terms have less that the long ago past has less influence than the recent
and less impact. past. Geographers assert that there occur comparable
decays of similarity in space. Thus, given spatial locations
Near Things (that is, latitudes and longitudes), one can to some ex-
tent predict (or estimate) the amount and types of in-
My use of the term near was equally ambiguous, as teraction, or the degrees of similarity, between these
pointed out by several of the discussants. Geographers locations. Is it more interesting when put the other way
(e.g., Gatrell 1983) have often studied distance in its around: given the degree of similarity, or interaction,
various forms. In addition to the metrical spherical (or between places, can one predict their locations (that is,
 Tobler: On the First Law of Geography: A Reply 307

the latitude and longitude coordinates) of the spatial laws to individuals—are useful. Weintraub (2002, 205)
origins? For example in 1962 W. Bunge suggested the use notes that acceptance of a theory is a social process but
of geographical central place theory to predict the loca- also (p. 185) writes ‘‘Understanding the world in which
tion of Mayan cites in Central America. Similarly, Susan Newton lived and made his contributions offers insight
Weinberg and I (Tobler and Weinberg 1971) predicted into the formation and acceptance of his contributions
the location of many pre-Hittite towns in Cappadocia; without denying the truth of his theories.’’ And Newton
this is an as yet unverified prediction. Philosophically, of spent much of his time on mystical and alchemical
course, prediction of this type is generally impossible, but, subjects. Does this detract from his ‘‘laws’’?
empirically, it is done. Gauss successfully predicted the In the present instance the development of analytical
location of Ceres, and Schliemann found Troy. geography and geographic information science has
In one sense, prediction is actually too easy: to- combined to produce a fertile ground for statements
morrow’s weather is likely to be like that of today. Both such as the first law of geography. My own background
temporal and spatial forecasting must be discounted for and mathematical interests also contributed, as Barnes
the effects of persistence. Is this another instance of the rightly observes.
first law of geography? Miller expands on the notions of ‘‘related’’ and ‘‘near.’’
In economics, agglomeration tendencies result in He does this in the context of spatial analysis. Here one
geographic concentrations of businesses that are related notes immediately the distinction observable between
to each other (Baldwin et al. 2003) and there are ‘‘spill the geographer as ‘‘natural scientist’’ and the individuals
over’’ situations (Palinck and Nijkamp 1975). The open- who are more interested in verstehen, a distinction nicely
ing of a shopping center generally results in stores spring- made by Catton (1965). The latter investigators hardly
ing up on the periphery. These are all manifestations of deal with autocorrelation or interpolation and similar
the ‘‘stickiness’’ of temporal and spatial effects. techniques. Nor do they use analytical computational
There is also obvious evidence that one must be methods in studying ‘‘nearness.’’ He, Miller, goes on to
carefully critical in applying the first law of geography. point out how nearness can be extended to include
Anisotropic effects do occur, and so do discontinuities. temporal relations, adding ‘‘when you are’’ to ‘‘where you
The world is not always regular and predictable. For are.’’ This seems like a useful concept. Certainly, the
example, topographic interpolation in the badlands of long-ago past influences us less than the recent past. The
South Dakota does not work well. The Mal Pais area in statistician recognizes this in the concept of an ‘‘auto-
New Mexico is also a region of rough topography. Per- regressive’’ model, also briefly explored in my paper
haps the name badlands (or Mal Pais) is in recognition of ‘‘Cellular Geography’’ (Tobler 1979).
the local failure of the first law of geography. In ‘‘Doing Justice to the Law’’ Phillips is concerned,
Is not much social strife, including urban riots and inter alia, with causal relations and takes as the point of
racial discrimination, due to a perceived incompatibility departure his field of physical geography. He then rightly
between neighbors? Spatial autocorrelation need not points out the many contingencies and conditions that
always be positive, as evidenced by the NIMBY con- must be attached to the statement of the first law of
troversy. When one wanders into a new environment—a geography. In my paper I asserted that I need not take
previously unexperienced part of a city, for example—it into account all of the activities throughout the world,
really does feel ‘‘strange.’’ Why is this? Can it explain a thus ignoring butterflies in Brazil, as well as population
part of the attraction of tourism? growth in that country and immigration from Brazil. Nor
do I know of any climate modelers who concern them-
selves with butterflies, even though the climatic system
The Discussion is multifaceted. Lewin’s ‘‘field theory’’ also excludes
‘‘events occurring at a remote distance’’ (1951, xii). On
To turn to the individual contributions, Barnes seems the whole I am in agreement with Phillips’s indication
to feel that the context of the discovery of a law some- of the complexity and contingent nature of our field of
how affects its validity. As anecdotal information, I have study. Most model assertions require the ceteris paribus
always found short historical commentary provides in- [everything else being equal] assumption.
sight and puts the human nature of science into per- I find that Smith fits quite well into Catton’s model of
spective but does not really change the result. Thus the the verstehen class of scholar, also emphasized by Giddens
validity would not be affected, even if the inventor’s and Flyvbjerg. He makes many good points, but I find it
(discoverer’s) name, or place in history, were not known. difficult to see even a remote connection between po-
As mnemonic shorthand the names—the attribution of pulation growth in the Detroit region to shoes and
308 Forum: On Tobler’s First Law of Geography

blisters. He deplores the debasement (perversion? evo- so. It is again a sparse matrix, and the permutation might
lution?) of language, but I do not find this particularly arrange the people by geographic distance. In spite of
important; rather, if true, it seems a natural human lin- this sparsity the ‘‘small-world’’ phenomena suggests that
guistic trait. everybody is only about six steps away from everybody
The panel discussant, Goodchild, takes a stance similar else (Milgram 1967). Is there a similar effect in the
to that of Miller, focussing on geographical information neurons in the brain? Perhaps this is a clue as to how
science, and the attendant analytical procedures. He goes the brain works, or how society works. What is left out
beyond this to speculate on further ‘‘laws,’’ making the here is the temporal variation of the connections.
important point that the first law, as constituted, deals One worm (caenorhabitis elegans) has been studied to
with a bivariate relation, a second-order effect. He then the point where its nervous system has been completely
considers other, perhaps more fundamental, possible mapped. It has 282 neurons and the topic of its inter-
‘‘principles’’ of geographic science. Spaces of social inter- connections—as many as 74 for one neuron—is actively
action, mentioned by Goodchild, are a much-studied, being pursued (Watts and Strogatz 1998).
dynamically growing endeavor, often using the multi- Similar to this, it is my understanding that when
dimensional scaling principle, which equates similarity one receptor in the human eye is activated nearby (i.e.,
with distance, but even allows the influence of distance to adjacent) receptors are inhibited. In both of these
vary from place to place, and varying by direction (Davies situations geometry plays an important part.
and Coxon 1982). His closing paragraphs remind me of a
statement found in Lösch (1954, 363), namely that the Conclusion
purpose of theory is ‘‘to test reality,’’ not the opposite, as is
often assumed. To summarize a long discussion, it can be seen that
Goodchild also introduces an interesting argument there are quite a number of references to laws relating
with respect to the physical environment, namely ‘‘what to human behavior. Just as my reference to things being
if [the first law] were not true?’’ He interprets this as related is not terribly novel, the assertion about near
meaning that there would be infinite local variation and, things being similar also has lots of precedents.5 Perhaps
consequently, complete unpredictability. Every spatial what is unique is that I put these two things together and
environment would become excessively chaotic. Does called the result a ‘‘law.’’ The fact that near things are
this also apply to social environments? more related than distant things seems a fundamental
He also raises the intriguing question of whether the property of geography and rather easily explained.
law can be extended to other spaces. Could it be that Finally, I would like to convey my thanks to Professor
the connectivity and interaction of neurons in the Sui for arranging this discussion panel and for chal-
human brain might exhibit properties similar to those lenging and stimulating questions concerning my reply.
observed in society? Within an order of magnitude, both The participants, Barnes, Miller, Phillips, Smith, and
seem to have the same number of components, and Goodchild, have all taken their obligation seriously.
both are characterized by relatively sparse connectivity Their contributions have led me to contemplate many
matrices and exhibit local interaction. What I mean by ideas and references with which I was previously un-
this is to suppose that you could label all of the human familiar. It also has provided a forum for a study of the
neurons and then list them in a table, with the source diversity of viewpoints within the field of geography. I did
neuron on the left margin and the destination neuron not expect such a discussion when I wrote that paper in
across the top. You would then have a table that is 1970. I was just having fun doing an animation in order
approximately 10 billion by 10 billion in size. Indicate by to bring time into geography more explicitly. Another
a one in the table all those neurons that send messages instance of the law of unexpected consequences?
directly to another neuron. Each neuron seems to
interact with about 1,000 others, so that the table is
asymmetric but mostly empty. Permute the table so that Notes
the interacting neurons lie near the diagonal, and the 1. Arbia’s second law of geography reads ‘‘Everything is related
sparseness of the table is then more easily seen. Are to everything else, but things observed at a coarse spatial
the neurons now ordered by their spatial separation? resolution are more related than things observed at a finer
Next do the same with all of the people in the world. resolution.’’ This suggests that aggregation has a smoothing
effect, as is well known (see Tobler 1969, 1990). My second
This conceptual table is approximately 6 billion by 6 law of geography asserts that ‘‘the phenomenon external to
billion in size. But most people routinely interact directly [a geographic] area of interest affects what goes on in the
with only a few others, perhaps also only a thousand or inside; a sufficiently common occurrence as to warrant being
 Tobler: On the First Law of Geography: A Reply 309

called the second law of geography.’’ This is comparable to Feynman, R. 1967. The character of physical law. Cambridge,
the need for ‘‘boundary conditions’’ in many physical pro- MA: MIT Press.
blems. It also relates, perhaps inversely, to Foucault’s (1979) Fisher, R. A. 1935. The design of experiments. Edinburgh: Oliver
emphasis on ‘‘confinement.’’ and Boyd.
2. Giddens (1984) also comments extensively on the rela- Flyvbjerg, B. 2001. Making social science matter (Sampson trans-
tion between geography and sociology. Concerning distance, lation). Cambridge: Cambridge University Press.
he states (p. 363), ’’Distance in space is apparently easy to Foucault, M. 1979. Discipline and punishment. Harmondsworth,
comprehend and to cope with conceptually’’ [sic]. He goes U.K.: Penguin.
on to say, ‘‘human beings do make their own geography.’’ He Gandin, L. 1963. Objective analysis of meteorological fields. GI-
finds great value in the concept of regionalization (p. xxv), MIZ, Leningrad (U.S. Department of Commerce transla-
but eschews the idea of laws in the social sciences (pp. xxxii– tion TT 65-50007).
xxxiv). Garfield, E. 1980. Bradford’s law and related statistical patterns.
3. The attribution of this statement to Chief Seattle has Essays of an information scientist IV, 476–83. Philadelphia:
been questioned. It is likely a myth. See J. L. Clark (1986), ISI Press.
‘‘Thus Spoke Chief Seattle: The Story of An Undocumented Gatrell, A. 1983. Distance and space. Oxford: Clarendon Press.
Speech,’’ Prologue, Spring 18(1). Also: http://www.jalcyon. Giddens, A. 1982. Hermeneutics and social theory. In Profiles
com/arborhts/chiefsea.html and http://www.webcom.com/ and critiques in social theory, A. Giddens, ed. Berkeley:
duane/seattle.html. University of California Press.
4. This phrase, ‘‘the law of unexpected consequences,’’ is quite FFF. 1984. The constitution of society. Berkeley: University of
common—as can be seen by entering it into Google on the California Press.
Internet—and apparently, quite well understood, though Golledge, R., and D. Amadeo. 1968. On laws in geography.
the origin is unclear. The distinguished historian H. Wayne Annals of the Association of American Geographers 58 (4):
Morgan, for example, uses it in his 1998 Brewster lecture 760–74.
in Greenville without his feeling any need for justification Goodchild, M., and D. Janelle. 1988. Specialization in the struc-
or elaboration. ture and organization of Geography. Annals of the Associa-
5. An anonymous reviewer suggested looking at a book by R. A. tion of American Geographers 78 (1): 1–28.
Fisher (1935). I did this and found the following on page 66: Hägerstrand, T. 1957. Migration and area. In Migration in Sweden,
‘‘the widely verified fact that patches in close proximity are Lund Studies in Geography, D. Hannerberg et al., Vol. 13,
commonly more alike, as judged by the yield of crops, than 17–128.
those which are further apart.’’ Hempel, C. 1996. The function of general laws in history. In
Logical empiricism and the special sciences, ed. H. Reich-
enbach, H. Feigl, and E. Nagel (Vol. 4 of Science and
References philosophy in the twentieth century, ed. S. Sarkar), 237–50.
New York: Garland.
Arbia, G., R. Benedetti, and G. Espa. 1996. Effects of the MAUP Hubert, L. 1974. Some applications of graph theory and related
on image classification. Geographical Systems 3:123–41. non-metric techniques to problems of approximate seria-
Baldwin, R., R. Forslid, P. Martin, G. Ottaviano, and F. Nicoud. tion. British Journal of Mathematical & Statistical Psychology
2003. Economic geography and public policy. Princeton, NJ: 27 (2): 133–54.
Princeton University Press. Lakatos, I., and P. Feyerabend. 1999. For and against method (ed.
Bunge, W. 1962. Theoretical geography, 2nd ed. Lund, M. Motterlini). Chicago: University of Chicago Press.
Sweden: Gleerup. Lewin, K. 1951. Field theory in social science (ed. D. Cartwright).
Cameron, J. 1997. Heart steps. New York: Putnam. New York: Harper Row.
Campbell, J. 1988. The power of myth: Sacrifice and bliss, video Lodge, D. 1984. Small world: An academic romance. New York:
with Bill Moyers. New York: Mystic Fire Video. Macmillan.
Catton, W. Jr. 1965. The concept of ‘‘mass’’ in the sociological Lösch, A. 1954. The economics of location (Woglom translation).
version of gravitation. In Mathematical explorations in beha- New Haven, CT: Yale University Press.
vioral science, ed. F. Massarik and P. Ratrosh, 287–316. Lotka, A. 1929. The frequency distribution of scientific producti-
Homewood, IL: R.D. Irwin. vity. Journal of the Washington Academy of Sciences 16:317–23.
Coxon, A. 1982. The user’s guide to multidimensional scaling. Mandelbrot, B. 1965. A class of long-tailed probability distribu-
Exeter, U.K.: Heinemann. tions and the empirical distribution of city sizes. In Mathe-
Davies, P., and A. Coxon, eds. 1982. Key texts in multidimensional matical explorations in behavioral science, ed. F. Massarik and
scaling. Exeter, U.K.: Heinemann. P. Ratrosh, 322–32. Homewood, IL: R.D. Irwin.
Dorigo, G., and W. Tobler. 1983. Push-pull migration laws. Merton, R. 1973. The sociology of science. Chicago: University of
Annals of the Association of American Geographers 73 (1): Chicago Press.
1–17. Milgram, S. 1967. The small world problem. Psychology Today
Estes, J. 1996. Chairman’s remarks. Keynote presentation to the 2:60–67.
International Steering Committee for Global Mapping, Morgan, W. 1998. Echoes and lessons from the Spanish Amer-
Tsubuka, Japan, Feb. 14 (available online at http://www. ican War. The Lawrence F. Brewster Lecture in History XVII,
iscgm.org/html4/index_c6_s1_ss1.html). East Carolina University, Greenville, North Carolina.
Fabrikant, S., M. Ruocco, R. Middleton, and D. Montello. 2002. Palinck, J., and P. Nijkamp. 1975. Operational theory and method
The first law of cognitive geography: Distance and simi- in regional analysis. Lexington, MA: Farnborough.
larity in semantic space. Proceedings of GIScience 2002. Palsky, G. 1996. Des chiffres et des cartes. Ministère de l’enseign-
Boulder, CO: 31–33. ment Supèrieur et de la Recherche, Paris.
310 Forum: On Tobler’s First Law of Geography

Patinkin, D. 1948. Relative prices, Say’s law, and the demand for FFF. 1990. Frame independent spatial analysis. In Accuracy
money. Econometrica 16:135–54. of spatial databases, ed. M. Goodchild and S. Gopal, 115–22.
Ravenstein, E. 1885. The laws of migration. Journal of the London: Taylor & Francis.
Statistical Society 28:167–235. FFF. 1995. Migration: Ravenstein, Thornthwaite, and be-
Reichenbach, H. 1952. The logical foundations of quantum yond. Urban Geography 16 (4): 327–43.
mechanics. In Logical empiricism and the special sciences, ed. FFF. 1999. Linear pycnophylatic reallocation—Comment
H. Reichenbach, H. Feigl, and E. Nagel (Vol. 4 of Science on a paper by D. Martin. International Journal of Geograph-
and philosophy in the twentieth century, ed. S. Sarkar), 55–95. ical Information Science 13 (1): 85–90.
New York: Garland. Tobler, W., and S. Weinberg. 1971. A Cappadocian speculation.
Simmons, G. 1992. Calculus gems. New York: McGraw-Hill. Nature 231 (5297): 39–42.
Sokal, A., and J. Bricmont. 1998. Fashionable nonsense: Postmo- Watts, D., and S. Strogatz. 1998. Collective dynamics of ‘‘small
dern intellectuals’ abuse of science. New York: Picador. world’’ networks. Nature 393:440–42.
Tobler, W. 1969. Geographical filters and their inverses. Geo- Weintraub, E. 2002. How economics became a mathematical
graphical Analysis 1 (3): 234–53. science. Durham, NC: Duke University Press.
FFF. 1979. Cellular geography. In Philosophy in geography, ed. Zipf, G. 1949. Human behavior and the principle of least effort.
S. Gale and G. Olsson, 379–86. Dortrecht: Riedel. Cambridge, MA: Addison-Wesley.

Correspondence: Geography Department, University of California, Santa Barbara, CA 93106-4060.